Today Google unveiled a series of new generative AI features across a range of its products including Gmail and its other Workspace apps, Google Docs, Sheets, and Slides. The AI-powered features will roll out to trusted testers in the coming weeks, and once they’ve been refined, Google says they will be available more generally. If this feels to you like Google is trying to play catch up with Microsoft and its multi-billion investment in/collaboration with OpenAI, the developers of ChatGPT and DALL-E 2, , well, you wouldn’t be wrong. According to The New York Times, it’s been a “code red” situation inside the company since ChatGPT launched last year, with plans to launch as many as 20 new products to address the perceived gap.
Still, in Google’s press releases, the company is quick to point out that it’s actually been doing this AI thing for a long time. A blog post from earlier this year lists nine ways that AI is used in the company’s products, including in Google Search, Maps, YouTube, Gmail, and, of course, Ads. Its existing AI features, like Smart Compose and Smart Reply in Gmail, are apparently already “helping 3 billion users.” And we can’t forget about the furore last year when an engineer got fired for claiming that LaMDA, a large language model, was sentient. It’s not Google who’s slow—it’s Microsoft, okay?
While Google announced a number of other features, it’s the generative AI integrations with apps like Gmail and Docs that are the most interesting. And assuming the beta testing goes well, they will likely be used by far more people.
According to Google, the new features will soon allow you to draft messages, reply to messages, summarize conversations, and prioritize messages in Gmail, brainstorm ideas, get your work proofread, generate text, and rewrite text in Docs, create AI-generated images, audio, and videos in Slides, capture notes and generate backgrounds in Meet, and more easily “go from raw data to insights and analysis” in Sheets (although these appear not to be connected to Google’s new AI chatbot, Bard).
In the blog post announcing the new features, Johanna Voolich Wright, vice president of product at Google Workspace, gives a few specific examples. In Docs, she shows the generative AI creating a rough draft of a job post for a regional sales rep, and in Gmail she shows it turning a short bulleted list into a formal email. Voolich Wright suggests these features would work whether you’re “a busy HR professional who needs to create customized job descriptions, or a parent drafting the invitation for your child’s pirate-themed birthday party.”
Voolich Wright is at pains to say that these features are meant to be you collaborating with AI, not letting it just do its own thing. “As we’ve experimented with generative AI ourselves, one thing is clear,” she writes. “AI is no replacement for the ingenuity, creativity, and smarts of real people.” In accordance with Google’s AI Principles, the generative AI is meant to do things like create first drafts that you edit and perfect, not publishable copy. You, the user, are meant to stay in control.
While these examples are cool and genuinely seem useful, all we have to go on right now is Google’s own announcement posts and demo videos. These tools aren’t yet available even to testers, so it’s important to treat the listed features and the examples Google gives with a bit of skepticism. We’re not saying that AI wasn’t used to generate the text in the demos Voolich Wright shows off, but they could just have easily been written by an intern in the marketing department as an example of what Google would like the new features to be able to do.
Still, Google has a legitimately world class AI research division and has been working on these sorts of features for more than six years. It might just be able to successfully integrate generative AI tools into some of its most popular products—and make them useful.